JP7067428B2 - Learning support devices, learning support methods and programs - Google Patents

Description

One aspect of the present invention relates to a learning support device, a learning support method, and a program that support a user's learning.

In learning systems such as online learning systems, a method of evaluating a user's learning ability is used in order to support the user's learning. As one of the methods for evaluating the learning ability, an item response theory is known in which the ability of the user and the difficulty level of the evaluation item are measured based on the response of the user to the evaluation item group.

According to item response theory, by simultaneously handling the characteristics (difficulty) of questions answered by multiple users (subjects) and the abilities (understanding or proficiency) of each user, it is not necessary to take the same test. , The degree of understanding of each user can be evaluated and compared among users (see, for example, Non-Patent Document 1). Further, there is also known a learning system that presents a problem of difficulty according to the estimated comprehension by using such a method of estimating comprehension of a user (see, for example, Non-Patent Document 2).

Yuki Tsukihara, Keiichi Suzuki, Hideo Hirose, "Mathematical Tests with Evaluation by Item Response Theory and Attempt to Implement in E-learning Systems", Computer & Education 24 (2008), 70-76 Chen, Chih-Ming, Hahn-Ming Lee, and Ya-Hui Chen. "Personalized e-learning system using item response theory." Computers & Education 44.3 (2005): 237-255

When item response theory is used, the user's ability and problem characteristics can be estimated by the following methods.

このモデルは、一般的に１パラメータロジスティック（１ＰＬ）モデル（ラッシュモデル）と呼ばれる。被験者母数θ_jは、各ユーザの能力値を表す（以下、「習熟度」と言う）。「項目」は、学習においては問題１問１問を指しており、項目母数ｂ_i は、各項目の難しさを表す（以下、「難易度」と言う）。 First, the probability P _ij for the user j to answer the item _i correctly can be expressed by the following equation 1 using the subject parameter θ _j and the item parameter bi.

This model is generally called a one-parameter logistic (1PL) model (rush model). The subject parameter θ _j represents the ability value of each user (hereinafter referred to as “proficiency level”). "Item" refers to one question and one question in learning, and the item parameter _bi represents the difficulty of each item (hereinafter referred to as "difficulty level").

Then, these parameters can be estimated from the likelihood of the following equation 2 when the reaction when the user j performs the item i is u _ij . Note that u _ij takes a value of 1 for a correct answer and 0 for an incorrect answer.

When both the proficiency level and the difficulty level are unknown, it is known that estimation can be performed using a simultaneous maximum likelihood estimation method, a peripheral maximum likelihood estimation method, a Bayesian estimation method, or the like.

By the way, in a system such as an online service teaching material distribution system in which each user freely selects and answers an item and obtains correct / incorrect data indicating the correctness of the answer, when the number of users or the number of items increases or decreases, it is new. It is necessary to calculate the proficiency level of the user and the difficulty level parameter for the newly added item. At this time, if the item response theory is calculated for all the data using the conventional technique, the scale of the proficiency level and the difficulty level will change every time the proficiency level and the difficulty level are estimated, and the calculation will be performed before. There is a problem (problem 0) that it cannot be compared with the parameters. For example, considering adding data to similar items later, if the added data has many correct answers (high correct answer rate), the difficulty level is estimated to be lower than before, and the proficiency level is also accompanied by it. It is easy to underestimate.

In order to solve this problem (problem 0), it is necessary to calculate the proficiency level of a new user and the difficulty level of a new item without changing the difficulty level calculated at the initial stage. Specifically, for a new user or an existing user who has new correct / incorrect data, it is possible to substitute the already calculated difficulty level into the above equation 2 and obtain the proficiency level by a method such as maximum likelihood estimation. be. Similarly, for a new item or an existing item in which new correct / incorrect data exists, it is also possible to substitute the already calculated proficiency level into Equation 2 and obtain the difficulty level by a method such as maximum likelihood estimation. However, with these methods alone, the following problems remain and the accuracy of the parameters cannot be guaranteed.

(Problem 1) Deterioration of reliability of proficiency estimation due to variation in reliability of difficulty
For items with a small number of answers, or items with too high or too low a correct answer rate even if many answers are given, the reliability of the difficulty level calculated by the item response theory becomes low. When estimating the proficiency level of a new user or an existing user who has new correct / incorrect data using the calculated difficulty level, it is newly estimated that the calculated difficulty level includes unreliable ones. The proficiency level is also low in reliability.

(Problem 2) Deterioration of reliability of difficulty estimation due to variation in reliability of proficiency
If the number of answers by the user is small, or if the correct answer rate is too high or too low, the reliability of the calculated proficiency level becomes low. Also, if it is assumed that there are users who will use the service for a long period of time, even if those users are proficient and the proficiency levels in the initial state and the final state are significantly different, a single user with the same proficiency level parameter. It must be treated as a user of, and even in such a case, the reliability of the proficiency level is considered to be low.

When estimating the difficulty level of a new item or an existing item that has new correct / incorrect data using the calculated proficiency level, it is newly estimated that the calculated proficiency level includes unreliable ones. The difficulty level to be done is also low in reliability.

(Problem 3) Insufficient data when the initial difficulty scale is fixed
Since the correct / incorrect data used for fixing the initial difficulty scale is a small amount of data accumulated at the initial stage of the service, the parameters estimated based on them may also be uncertain estimates. If necessary, it may be desirable to update each parameter to an estimate with a lot of correctness data.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a learning support device, a learning support method, and a program that enable highly reliable parameter estimation.

In order to solve the above problem, the first aspect of the present invention is a learning support device that supports a user's learning, and is correct / incorrect data representing the correctness of each answer of a plurality of users to a plurality of items to be learned. Based on the correct / incorrect data acquisition unit and the acquired correct / incorrect data, the proficiency level indicating the high ability of each user, the difficulty level indicating the difficulty of each item, and the proficiency level described above. And a parameter calculation unit that calculates the reliability indicating the high reliability for each of the above difficulty levels, and generates a trigger instructing to update at least one of the calculated proficiency levels or the difficulty levels. By using the trigger generator and the proficiency level or difficulty level having a reliability exceeding the first threshold value among the calculated proficiency levels or difficulty levels according to the generated trigger as a reference parameter. It is provided with an update control unit that recalculates at least one of the proficiency level or the difficulty level instructed to update by the above trigger.

In the second aspect of the present invention, in the learning support device of the first aspect, the number of correct / incorrect data in which the trigger generation unit is not used for calculating the proficiency level for a specific user exceeds the first number. When, when the instruction to update the proficiency level is received for the specific user, or when the reliability of the proficiency level calculated for the specific user is lower than the second threshold value and the proficiency level is proficient for the specific user. When there is correct / incorrect data that is not used for calculating the degree, a proficiency level update trigger that instructs the update of the proficiency level of the specific user is generated.

In the third aspect of the present invention, in the learning support device of the first aspect, the number of correct / incorrect data in which the trigger generation unit is not used for calculating the difficulty level for a specific item exceeds the second number. When the instruction to update the difficulty level for the specific item is received, or the reliability of the calculated difficulty level for the specific item is lower than the third threshold value and the difficulty level for the specific item is difficult. When there is correct / incorrect data that is not used for calculating the degree, a difficulty update trigger that instructs the update of the difficulty of the above specific item is generated.

A fourth aspect of the present invention is that in the learning support device of the first aspect, the cumulative number of correct / incorrect data acquired by the trigger generation unit by the correct / incorrect data acquisition unit is a natural number multiple of the third number. When it reaches, when the instruction to update the proficiency level and the difficulty level at the same time is received, the average value of the reliability level of the proficiency level for all users or the average value of the reliability level of the difficulty level for all items is the fourth. When it is lower than the threshold value, the ratio of the reliability of the proficiency level for all users or the reliability level of the difficulty level for all items, which is lower than the fifth threshold value, exceeds the sixth threshold value. Correct answer calculated based on the relational expression based on the correct answer rate calculated from the correct / incorrect data acquired by the correct / incorrect data acquisition unit and the proficiency level and difficulty level calculated by the parameter calculation unit. When the match rate with the rate is lower than the seventh threshold value, a simultaneous update trigger instructing simultaneous update of the proficiency level and the difficulty level is generated.

A fifth aspect of the present invention is associated with the specific user in the learning support device of the first aspect when the update control unit instructs the update of the proficiency level of the specific user. A new proficiency level is calculated based on the correct / incorrect data that has not been used to calculate the proficiency level among the correct / incorrect data having the user identification information, and the difference between the new proficiency level and the previously calculated proficiency level is When it is larger than the eighth threshold value, new user identification information different from the user identification information of the specific user is given to the new proficiency level and the correctness data used for the calculation. be.

A sixth aspect of the present invention is that in the learning support device of the first aspect, the update control unit determines the recalculated proficiency level or difficulty level and the previously calculated proficiency level or difficulty level, respectively. Update the proficiency or difficulty by multiplying by a weighting factor or by taking a moving average based on the recalculated proficiency or difficulty and the previously calculated proficiency or difficulty. I tried to do it.

According to the first aspect of the present invention, the proficiency level and the reliability of each user and the difficulty level and the reliability level of each item are calculated based on the acquired correctness data, and the proficiency level or the difficulty level is calculated. In response to the trigger instructing to update at least one of them, the parameter instructed to update by the trigger is recalculated using only the highly reliable parameter as the reference parameter. As a result, when calculating the item difficulty level and user proficiency level evaluated with the same difficulty level as the initially calculated difficulty level, the reliability evaluation of the parameters is used to make the data for each user and item dense. By removing the accompanying uncertain parameters, more probable parameters can be calculated.

That is, for the above-mentioned task 1, a new user or a new user or a new user or a new person by removing the unreliable difficulty level by the reliability evaluation of the calculated difficulty level and using only the highly reliable difficulty level as a reference parameter. It is possible to realize highly reliable proficiency estimation for existing users who have correct / incorrect data.

Further, regarding the above-mentioned task 2, a new item or a new item or a new item is obtained by removing the unreliable proficiency level by the reliability evaluation of the calculated proficiency level and using only the highly reliable proficiency level as a reference parameter. It is possible to realize a highly reliable difficulty estimation for an existing item in which correct / incorrect data of the above exists.

According to the second aspect of the present invention, when the correct / incorrect data of a new user is obtained or the correct / incorrect data of an existing user increases, or when a request from a user, a system administrator, or the like or a scheduled update instruction is received. Alternatively, when the proficiency level of a particular user is unreliable and there is new correctness data for that particular user, a trigger is generated instructing the proficiency level to be updated. This makes it possible to appropriately generate and output a trigger instructing the update of the proficiency level in a situation where the proficiency level needs to be updated.

According to the third aspect of the present invention, when the correct / incorrect data of a new item is obtained or the correct / incorrect data of an existing item is increased, a request from a user, a system administrator, or the like or a scheduled update instruction is accepted. When, or when the difficulty level of a particular item is unreliable and there is new correctness data for that particular item, a trigger is generated instructing the difficulty level to be updated. As a result, in a situation where the difficulty level needs to be updated, a trigger instructing the difficulty level update can be appropriately generated and output.

According to the fourth aspect of the present invention, when the cumulative number of acquired correct / incorrect data increases, when a request from a user, a system administrator, or the like or a scheduled update instruction is received, the calculated difficulty level or mastery is received. When the average value of reliability is low, the difficulty level with low reliability or the percentage of the total proficiency level increases, the correct answer rate obtained from the actual correctness data and the calculated parameters are used to determine the degree. When the match rate with the correct answer rate obtained from the relational expression is low, a trigger is generated instructing the proficiency level and the difficulty level to be updated at the same time. As a result, in a situation where simultaneous update of proficiency level and difficulty level is required, it is possible to appropriately generate and output a trigger instructing simultaneous update of proficiency level and difficulty level.

According to the fifth aspect of the present invention, a new proficiency level is calculated for a specific user by using correct / incorrect data that has not been used for calculating the proficiency level in response to a trigger instructing the update of the proficiency level. Occasionally, when the difference between the new proficiency level and the previous proficiency level is large, the new proficiency level and the correct / incorrect data used for the calculation include user identification information different from the user identification information of the specific user. Granted. When there are long-term users or when there is a large change in the user's proficiency, it is difficult to express the user's characteristics with a single proficiency, and the fit between the model and the actual data becomes poor. However, by the above method, the data of a single user is actually the data, but the user's proficiency level changes by treating the data as if it were the data of another user with different proficiency levels. It is possible to reduce the discrepancy between the model and the data and improve the estimation accuracy of each parameter while estimating the proficiency level according to the actual situation of.

According to a sixth aspect of the present invention, the proficiency or difficulty is recalculated using reliable reference parameters, and predetermined weights are given to each of the newly calculated parameter and the previously calculated parameter. New update parameters are obtained by multiplying by the coefficients and then adding, or by taking the moving average of the newly calculated parameters and the previously calculated parameters. As a result, instead of using the newly calculated proficiency level or difficulty level as it is, the update is performed after performing an operation considering the change over time with respect to the previously calculated proficiency level or difficulty level. Therefore, even when data for a new user or a new item is generated, the difficulty level and proficiency level can be calculated without being inconsistent with the previous difficulty level scale, and the proficiency level and item difficulty level are presented to the user. In that case, it can be used without discomfort even when compared with the parameters calculated before new data is generated.

That is, according to each aspect of the present invention, it is possible to provide a learning support device, a learning support method, and a program that enable highly reliable parameter estimation.

FIG. 1 is a block diagram showing a functional configuration of a learning support device according to an embodiment of the present invention. FIG. 2 is a block diagram showing a detailed configuration of a proficiency level update control unit among the functional units included in the learning support device shown in FIG. FIG. 3 is a block diagram showing a detailed configuration of a difficulty level update control unit among the functional units included in the learning support device shown in FIG. 1. FIG. 4 is a block diagram showing a detailed configuration of a simultaneous update control unit among the functional units included in the learning support device shown in FIG. 1. FIG. 5 is a flowchart showing a processing procedure of initial parameter calculation by the learning support device shown in FIG. 1 and an example of the processing content. FIG. 6 is a flowchart showing a processing procedure for updating the proficiency level by the learning support device shown in FIG. 1 and a first example of the processing content. FIG. 7 is a flowchart showing a processing procedure for updating the proficiency level by the learning support device shown in FIG. 1 and a second example of the processing content. FIG. 8 is a flowchart showing a processing procedure for updating the difficulty level by the learning support device shown in FIG. 1 and an example of the processing content. FIG. 9 is a flowchart showing a processing procedure for simultaneous update of proficiency level and difficulty level by the learning support device shown in FIG. 1 and an example of the processing content. FIG. 10 is a diagram showing an example of correct / incorrect data. FIG. 11 is a diagram showing an example of proficiency level / reliability data. FIG. 12 is a diagram showing an example of difficulty / reliability data. FIG. 13 is a diagram showing an example of an update trigger table that defines an update instruction trigger.

Hereinafter, embodiments relating to the present invention will be described with reference to the drawings.

[One Embodiment]
I. Configuration (1) Overall Configuration FIG. 1 is a block diagram showing a functional configuration of the learning support device 1 according to the embodiment of the present invention.

The learning support device 1 is used, for example, as a part of an online learning system provided by an educational institution such as a university or an educational service provider, and is configured by, for example, a server computer or a personal computer. The learning support device 1 acquires and acquires correct / incorrect data indicating the correctness (correct answer or incorrect answer) of the answers of a plurality of users to a plurality of problems (items) via a communication network or an input device. By calculating various parameters based on the obtained correct / incorrect data and outputting the calculated parameters as appropriate, it is possible to support the learning of the user.

The learning support device 1 according to one embodiment includes an input / output interface unit 10, a control unit 20, and a storage unit 30.

The input / output interface unit 10 includes, for example, one or more wired or wireless communication interface units, and enables information to be transmitted / received to / from an external device. As the wired interface, for example, a wired LAN is used, and as the wireless interface, an interface adopting a low power wireless data communication standard such as a wireless LAN or Bluetooth (registered trademark) is used.

For example, the learning support device 1 takes in the correct / incorrect data collected by the online learning system via the input / output interface unit 10 at a fixed amount or at regular intervals, and performs a process of passing the taken-in data to the control unit 20. The input / output interface unit 10 also performs a process of taking in an instruction or the like input through an input device (not shown) such as a keyboard and passing it to the control unit 20. Further, the input / output interface unit 10 outputs the parameter calculation result output from the control unit 20 to a display device (not shown) for presentation to the user or the operator, or transmits the parameter calculation result to an external device via the network. It is also possible to perform the processing to be performed.

The storage unit 30 uses, for example, a non-volatile memory such as an HDD (Hard Disk Drive) or SSD (Solid State Drive) that can be written and read at any time as a storage medium, and in order to realize this embodiment. In addition to the program storage unit, the correct / incorrect data storage unit 31, the parameter storage unit 32, and the trigger table storage unit 33 are provided as the storage areas required for the above.

The correct / incorrect data storage unit 31 stores the correct / incorrect data table for storing the correct / incorrect data acquired via the input / output interface unit 10.

The parameter storage unit 32 stores various parameters calculated based on the acquired correct / incorrect data. The parameter storage unit is particularly used to store a proficiency level table that stores the calculated proficiency level and its reliability, and a difficulty level table that stores the calculated difficulty level and its reliability.

The trigger table storage unit 33 stores an update trigger table that defines an update trigger.

However, the storage units 31 to 33 are not indispensable configurations, and the learning support device 1 may acquire necessary data from an external device at any time. Alternatively, the storage units 31 to 33 do not have to be built in the learning support device 1, and are provided in, for example, an external storage medium such as a USB memory or a storage device such as a database server arranged in the cloud. It may be the one that was given.

The control unit 20 has a hardware processor such as a CPU (Central Processing Unit) or MPU (Micro Processing Unit) (not shown) and a memory such as a DRAM (Dynamic Random Access Memory) or a SRAM (Static Random Access Memory). The processing functions required to implement this embodiment include the correct / incorrect data acquisition unit 21, the parameter calculation unit 22, the parameter output control unit 23, the update trigger determination unit 24, the trigger output control unit 25, and the update control. It is equipped with a unit 26. All of these processing functions are realized by causing the processor to execute the program stored in the storage unit 30. The control unit 20 may also be implemented in a variety of other forms, including integrated circuits such as ASICs (Application Specific Integrated Circuits) and FPGAs (field-programmable gate arrays).

The correct / incorrect data acquisition unit 21 acquires correct / incorrect data via the input / output interface unit 10 and stores the correct / incorrect data in the correct / incorrect data storage unit 31.

The parameter calculation unit 22 reads out the correct / incorrect data stored in the correct / incorrect data storage unit 31, calculates various parameters based on the read correct / incorrect data, and stores the calculated parameters in the parameter storage unit 32. .. In one embodiment, the parameter calculation unit 22 calculates and further calculates the proficiency level indicating the high ability of each user and the difficulty level indicating the difficulty of each item by using the above-mentioned item response theory. Calculate the reliability that indicates the high degree of reliability for each of the proficiency level and the difficulty level.

The parameter output control unit 23 creates output data based on the parameters stored in the parameter storage unit 32, and performs a process of outputting the data to a display device or an external device via the input / output interface unit 10. For example, the parameter output control unit 23 may generate and output output data including the difficulty level of each item and the proficiency level of each user in response to a request from an external application such as a learning system or a musical instrument performance support system. can.

The update trigger determination unit 24 determines whether or not the calculated parameter stored in the parameter storage unit 32 needs to be updated, and writes the determination result in the trigger table stored in the trigger table storage unit 33. conduct.

The trigger output control unit 25 refers to the trigger table stored in the trigger table storage unit 33, generates a necessary update trigger, and outputs the necessary update trigger to the update control unit 26.

The update control unit 26 receives the update trigger output from the trigger output control unit 25 and performs the update control process according to the received update trigger. The proficiency level update control unit 261 and the difficulty level update control unit 262. And a simultaneous update control unit 263 is provided.

The proficiency level update control unit 261 receives correct / incorrect data stored in the correct / incorrect data storage unit 31 and a parameter storage unit 32 in response to the proficiency level update trigger instructing the update of the proficiency level output from the trigger output control unit 25. Based on the difficulty level stored in the parameter storage unit 32, a process of calculating the proficiency level based on the difficulty level stored in the parameter storage unit 32 is performed. Here, the difficulty scale represents a scale of where the difficulty is located when, for example, a difficulty range from -4 to +4 is set in order to compare the difficulty between items.

The difficulty level update control unit 262 receives the correct / incorrect data stored in the correct / incorrect data storage unit 31 and the parameter storage unit 32 in response to the difficulty update trigger instructing the update of the difficulty level output from the trigger output control unit 25. Based on the proficiency level stored in the parameter storage unit 32, a process of calculating the difficulty level based on the difficulty level stored in the parameter storage unit 32 is performed.

The simultaneous update control unit 263 stores correct / incorrect data stored in the correct / incorrect data storage unit 31 and parameter storage in response to the simultaneous update trigger instructing simultaneous update of the proficiency level and the difficulty level output from the trigger output control unit 25. A process of calculating the proficiency level and the difficulty level is performed based on the parameters stored in the unit 32.

(2) Configuration of Proficiency Update Control Unit FIG. 2 shows a more detailed functional configuration of the proficiency level update control unit 261 included in the learning support device 1 configured as described above. The proficiency level update control unit 261 calculates the proficiency level in response to the proficiency level update trigger and outputs the calculated proficiency level. The data extraction unit 2611, the reference data selection unit 2612, and the proficiency level / reliability It includes a degree calculation unit 2613 and a reliability determination unit 2614.

The data extraction unit 2611 is a process of extracting correct / incorrect data having the user ID from the correct / incorrect data storage unit 31 based on the user identification information (user ID) that specifies the update target, which is included as additional information in the proficiency level update trigger. I do.

The reference data selection unit 2612 performs a process of selecting reference data to be used for calculating the proficiency level from the extracted correct / incorrect data. For the selection of the reference data, the difficulty level stored in the parameter storage unit 32 and its reliability are used. As a result of selection, it is possible to use all the data.

The proficiency level / reliability calculation unit 2613 calculates the proficiency level of the target user using the reference data selected by the reference data selection unit 2612, and the proficiency level is calculated from the calculated proficiency level and the correct / incorrect data used for the calculation. Performs the process of calculating the reliability of the degree.

The reliability determination unit 2614 performs a process of comparing the calculated reliability with the preset threshold value. When the calculated reliability is equal to or higher than the preset threshold value, the reliability determination unit 2614 outputs the calculated proficiency level together with the reliability in order to update the data stored in the parameter storage unit 32. If the calculated reliability is less than the preset threshold, the reliability determination unit 2614 does not output the proficiency level. At this time, the reliability determination unit 2614 can instruct the reference data selection unit 2612 to change the reference data selection criteria.

(3) Configuration of Difficulty Update Control Unit FIG. 3 shows a more detailed functional configuration of the difficulty update control unit 262 included in the learning support device 1 configured as described above. The difficulty level update control unit 262 calculates the difficulty level in response to the difficulty level update trigger and outputs the calculated difficulty level. The data extraction unit 2621, the reference data selection unit 2622, and the difficulty level / reliability It includes a degree calculation unit 2623 and a reliability determination unit 2624.

The data extraction unit 2621 is a process of extracting correct / incorrect data having the item ID from the correct / incorrect data storage unit 31 based on the item identification information (item ID) that specifies the update target, which is included as additional information in the difficulty update trigger. I do.

The reference data selection unit 2622 performs a process of selecting reference data to be used for calculating the difficulty level from the extracted correct / incorrect data. For the selection of the reference data, the proficiency level stored in the parameter storage unit 32 and its reliability are used. As a result of selection, it is possible to use all the data.

The difficulty / reliability calculation unit 2623 calculates the difficulty level of the target item using the reference data selected by the reference data selection unit 2622, and the difficulty level is calculated from the calculated difficulty level and the correct / incorrect data used for the calculation. Performs the process of calculating the reliability of the degree.

The reliability determination unit 2624 performs a process of comparing the calculated reliability with the preset threshold value. When the calculated reliability is equal to or higher than the preset threshold value, the reliability determination unit 2624 outputs the calculated difficulty level together with the reliability in order to update the data stored in the parameter storage unit 32. If the calculated reliability is less than the preset threshold, the reliability determination unit 2624 does not output the difficulty level. At this time, the reliability determination unit 2624 can instruct the reference data selection unit 2622 to change the reference data selection criteria.

(4) Configuration of Simultaneous Update Control Unit FIG. 4 shows a more detailed functional configuration of the simultaneous update control unit 263 included in the learning support device 1 configured as described above. The simultaneous update control unit 263 calculates the proficiency level and the difficulty level in response to the simultaneous update trigger, and outputs the calculated proficiency level and the difficulty level. The reference data selection unit 2631, the proficiency level / difficulty level, A reliability calculation unit 2632 and a reliability determination unit 2633 are provided.

The reference data selection unit 2631 selects reference data based on the total correctness data stored in the correct / incorrect data storage unit 31, the proficiency level, the difficulty level, and the reliability of each stored in the parameter storage unit 32. Perform processing.

The proficiency / difficulty / reliability calculation unit 2632 calculates a new proficiency level, difficulty level, and reliability of each using the reference data selected by the reference data selection unit 2631.

The reliability determination unit 2633 performs a process of comparing the calculated reliability with the preset threshold value. When the calculated reliability is equal to or higher than the preset threshold value, the reliability determination unit 2633 updates the data stored in the parameter storage unit 32, so that the calculated proficiency level and the difficulty level are each trusted. Output with degree.

II. Operation Next, the processing operation by the learning support device 1 configured as described above will be described.

(1) Initial parameter calculation process First, the learning support device 1 acquires initial data, calculates the proficiency level and the difficulty level from the initial data, and performs a process of fixing the difficulty level scale. The initial data is acquired via a learning system or the like according to the user's use of the service, and includes correct / incorrect data indicating the correctness of the answers of the plurality of users to the plurality of problems. FIG. 5 is a flowchart showing a processing procedure and processing contents for calculating the proficiency level and the difficulty level from such initial data.

The learning support device 1 monitors the presence / absence of data input under the control of the control unit 20. In this state, for example, correct / incorrect data is collected via the learning system, and when the collected correct / incorrect data is transmitted to the learning support device 1, the learning support device 1 detects the reception of this data and performs the following processing. Start.

First, in step S101, the learning support device 1 acquires correct / incorrect data via the input / output interface unit 10 under the control of the correct / incorrect data acquisition unit 21, and stores the correct / incorrect data in the correct / incorrect data storage unit 31.

FIG. 10 shows an example of a correct / incorrect data table stored in the correct / incorrect data storage unit 31. In the example of FIG. 10, in the correct / incorrect data table, the correct / incorrect data ID for identifying the data record, the user ID for identifying the user, the item ID for identifying the problem, and the correct / incorrect information indicating correct / incorrect answer (1 is correct answer, 0 is incorrect). Represents the answer), and contains a column of date and time information.

After acquiring and accumulating initial correctness data, the learning support device 1 receives input of a parameter calculation instruction from, for example, a user, a system administrator, or the like, and performs subsequent parameter calculation processing.

In step S102, the learning support device 1 reads the correct / incorrect data accumulated from the correct / incorrect data storage unit 31 under the control of the parameter calculation unit 22, and calculates the proficiency level and the difficulty level by batch processing. Peripheral maximum likelihood estimation, item response theory using Bayesian estimation, or other methods can be used to calculate proficiency and difficulty.

式中、Ｄは任意の定数であり、この実施形態では、Ｄ＝１．７である。 In step S103, the learning support device 1 calculates the reliability for each of the calculated proficiency level and the difficulty level under the control of the parameter calculation unit 22. The amount of test information and the number of answers can be used as an index of reliability. The amount of test information for the proficiency level and the difficulty level can be calculated using the following formulas 3 and 4.

In the equation, D is an arbitrary constant, and in this embodiment, D = 1.7.

In addition to the amount of test information, an index such as the Monte Carlo method can also be used as an index of reliability. In the case of the Monte Carlo method, the trial to randomly extract the correct / incorrect data corresponding to a certain item is repeated N times, the difficulty level is calculated from those samples, and the reciprocal of the variance of the calculated difficulty level distribution is used as the reliability. be able to. Similarly, for the proficiency level, the trial of randomly extracting the correct / incorrect data corresponding to a certain user is repeated N times, the proficiency level is calculated from those samples, and the reciprocal of the variance of the calculated proficiency level distribution is the reciprocal. Can be.

Next, in step S104, the learning support device 1 stores the proficiency level, the difficulty level, and the reliability of each calculated in step S102 and step S103 in the parameter storage unit 32 under the control of the parameter calculation unit 22.

If the difficulty level is set in advance before executing the above processing, the difficulty level can also be used. In this case, using the same method as the proficiency level update process described later, the proficiency level is calculated using the preset difficulty level, the reliability for each calculated proficiency level is calculated, and the proficiency level and difficulty level are calculated. The degree and reliability are stored in the parameter storage unit 32.

FIG. 11 shows an example of a proficiency level table that stores the proficiency level and its reliability stored in the parameter storage unit 32. In the example of FIG. 11, the proficiency level calculated for each user and the reliability thereof, and the correct / incorrect data ID used for calculating the proficiency level are stored in association with the user ID that identifies each user. The user ID "2_2" here means a second ID for the same user as the user ID "2". This will be further described in the following (3-1-2) regarding the proficiency level update process for long-term users.

Similarly, FIG. 12 shows an example of a difficulty level table that stores the difficulty level and the reliability thereof stored in the parameter storage unit 32. In the example of FIG. 12, the difficulty level calculated for each item, the reliability thereof, and the correct / incorrect data ID used for calculating the difficulty level are stored in association with the item ID that identifies each item.

(2) Generation / Output Process of Update Trigger Next, a process in which the learning support device 1 determines whether or not the parameter needs to be updated and outputs the update trigger will be described.

Under the control of the update trigger determination unit 24, the learning support device 1 refers to the data of the correct / incorrect data storage unit 31 and the parameter storage unit 32, for example, at regular time intervals, and determines whether or not the parameter needs to be updated. Then, the determination result is written in the trigger table in the trigger table storage unit 33.

FIG. 13 shows an example of an update trigger table that defines an update trigger stored in the trigger table storage unit 33. The update trigger includes three types: proficiency level update trigger, difficulty level update trigger, and simultaneous update trigger. When the update trigger determination unit 24 determines that the parameter needs to be updated, it inputs 1 to the ON / OFF column of the update trigger table and describes additional information as necessary. If the update is not required, enter 0 in the ON / OFF column to delete the additional information.

Further, the learning support device 1 monitors the update trigger table stored in the trigger table storage unit 33 under the control of the trigger output control unit 25. When the proficiency level update trigger is turned ON in the update trigger table, the trigger output control unit 25 generates a proficiency level update trigger, and the proficiency level update trigger and the user ID for which the proficiency level update should be performed are transmitted to the update control unit 26. Notice. When the difficulty level update trigger is turned ON, the trigger output control unit 25 generates a difficulty level update trigger, and notifies the update control unit 26 of the item ID for which the difficulty level update should be performed together with the difficulty level update trigger. When the simultaneous update trigger is ON, the trigger output control unit 25 generates a simultaneous update trigger and notifies the update control unit 26. The simultaneous update trigger does not require the update process at the same timing of the proficiency level and the difficulty level, but refers to a trigger that instructs the update of both the proficiency level and the difficulty level within a certain period of time.

Hereinafter, the determination of the necessity of parameter update in the update trigger determination unit 24 will be further described for each of the three types of update triggers.

(2-1) Proficiency update trigger A large number of examples can be considered as a method for determining the necessity of updating the proficiency parameter in the update trigger determination unit 24. Some of them are described below.

(2-1-1) When a new user is generated When it is determined that a new user has been generated, the update trigger determination unit 24 determines that the proficiency level needs to be updated for the new user.

For example, the update trigger determination unit 24 determines in the correct / incorrect data storage unit 31 whether or not there is N1 or more data of the user ID that is not stored in the proficiency level table of the parameter storage unit 32. When it is determined that N1 or more exist, the update trigger determination unit 24 determines that the proficiency level needs to be updated for the new user, and sets 1 in the ON / OFF column of the proficiency level update trigger in the trigger table to the additional information column. Enter the user ID. Note that N1 is an arbitrary integer of 1 or more.

(2-1-2) When the data of an existing user increases When the update trigger determination unit 24 determines that the data of an existing user has increased, it determines that the proficiency level of the user needs to be updated. ..

For example, the update trigger determination unit 24 is a user ID stored in the proficiency level table of the parameter storage unit 32 in the correct / incorrect data storage unit 31, and is a correct / incorrect data ID used for calculating the proficiency level table of the parameter storage unit 32. It is determined whether or not there are N2 or more correct / incorrect data not listed in the column. When it is determined that N2 or more exist, the update trigger determination unit 24 determines that the proficiency level needs to be updated for the user because the data of the existing user has increased, and the ON / OFF column of the proficiency level update trigger in the trigger table. 1 and the user ID in the additional information column. Note that N2 is an arbitrary integer of 1 or more, and is preferably the same as or different from N1.

(2-1-3) When an update trigger input is accepted The update trigger determination unit 24 determines that the proficiency level needs to be updated according to the acceptance of the input instructing the update.

For example, when an internal / external proficiency update instruction is received, which is generated according to a preset schedule or input via an input unit (not shown) when the user wants to recalculate his / her proficiency level. The update trigger determination unit 24 determines that the proficiency level needs to be updated for the user, and describes 1 in the ON / OFF column of the proficiency level update trigger in the trigger table and the user ID in the additional information column.

(2-1-4) When the reliability is low The update trigger determination unit 24 determines that the proficiency level needs to be updated for the corresponding user when it is determined that the reliability of the stored data is low.

For example, in the update trigger determination unit 24, the reliability of the proficiency level stored in the proficiency level table stored in the parameter storage unit 32 is lower than the preset threshold value T1, and the correct / incorrect data storage unit 31 stores. If the stored correct / incorrect data table contains correct / incorrect data that is not described as the correct / incorrect data ID used for calculating the proficiency level in the proficiency level table, it is determined that the proficiency level needs to be updated. The update trigger determination unit 24 describes 1 in the ON / OFF column of the proficiency level update trigger of the update trigger table, and the user ID of the corresponding user in the additional information column.

(2-2) Difficulty level update trigger A large number of examples can be considered as a method for determining the necessity of updating the difficulty level parameter in the update trigger determination unit 24. Some of them are described below.

(2-2-1) When a new item occurs When it is determined that a new item has occurred, the update trigger determination unit 24 determines that the difficulty level needs to be updated for the item.

For example, the update trigger determination unit 24 determines in the correct / incorrect data storage unit 31 whether or not there are N3 or more data of item IDs that are not stored in the difficulty table of the parameter storage unit 32. When it is determined that N3 or more exist, the update trigger determination unit 24 determines that the difficulty level needs to be updated for the new item, and sets 1 in the ON / OFF column of the difficulty level update trigger in the trigger table, and the additional information column. Enter the item ID in. N3 is an arbitrary integer of 1 or more.

(2-2-2) When the data of an existing item increases When the update trigger determination unit 24 determines that the data of an existing item has increased, it determines that the difficulty level needs to be updated for the item.

For example, the update trigger determination unit 24 is an item ID stored in the difficulty table of the parameter storage unit 32 in the correct / incorrect data storage unit 31, and is the correct / incorrect data ID used for calculating the difficulty table of the parameter storage unit 32. It is determined whether or not there are N4 or more correct / incorrect data not listed in the column. When it is determined that N4 or more exist, the update trigger determination unit 24 determines that the difficulty level needs to be updated for the item because the data of the existing item has increased, and turns ON / OFF the difficulty level update trigger of the trigger table. Enter 1 in the column and the item ID in the additional information column. Note that N4 is an arbitrary integer of 1 or more, and is preferably the same as or different from N3.

(2-2-3) When an update trigger input is accepted The update trigger determination unit 24 determines that the difficulty level needs to be updated according to the acceptance of the input instructing the update.

For example, it accepts difficulty update instructions from inside and outside that are generated according to a preset schedule or entered via an input unit (not shown) when the system administrator wants to recalculate the difficulty level of a specific item. At that time, the update trigger determination unit 24 determines that the difficulty level needs to be updated for the item, and describes 1 in the ON / OFF column of the difficulty level update trigger in the trigger table and the item ID in the additional information column.

(2-2-4) When the reliability is low When the update trigger determination unit 24 determines that the reliability of the stored data is low, it determines that the difficulty level needs to be updated for the corresponding item.

For example, in the update trigger determination unit 24, the reliability of the difficulty level stored in the difficulty level table stored in the parameter storage unit 32 is lower than the preset threshold value T2, and the correct / incorrect data storage unit 31 stores. If the stored correct / incorrect data table contains correct / incorrect data that is not described as the correct / incorrect data ID used for calculating the difficulty level in the difficulty level table, it is determined that the difficulty level needs to be updated. The update trigger determination unit 24 describes 1 in the ON / OFF column of the difficulty level update trigger of the update trigger table, and describes the item ID in the additional information column.

(2-3) Simultaneous update trigger A large number of examples can be considered as a method for determining the necessity of updating the proficiency level and the difficulty level in the update trigger determination unit 24. Some of them are described below.

(2-3-1) When the number of correct / incorrect data increases The update trigger determination unit 24 determines that the proficiency level and the difficulty level need to be updated every time the cumulative number of acquired correct / incorrect data exceeds a certain value. ..

For example, the update trigger determination unit 24 determines that when it is determined that the correct / incorrect data stored in the correct / incorrect data storage unit 31 has reached a multiple of M, it is necessary to update the proficiency level and the difficulty level, and the simultaneous update trigger 1 is described in the ON / OFF column. In addition, M is an arbitrary integer, and it is desirable that it is a large unit such as 10,000 or 100,000.

(2-3-2) When an update trigger input is accepted The update trigger determination unit 24 determines that the proficiency level and the difficulty level need to be updated according to the acceptance of the input instructing the update.

For example, when the system administrator wants to update the difficulty level, or when the update trigger determination unit 24 receives an instruction for simultaneous update from inside or outside, such as when updating according to a yearly or monthly schedule, the update trigger determination unit 24 receives an instruction for simultaneous update. The update trigger determination unit 24 describes 1 in the ON / OFF column for the simultaneous update trigger of the update trigger table.

(2-3-3) When the number of low-reliability items / users increases The update trigger determination unit 24 determines that the proficiency level and difficulty level need to be updated when it is determined that the number of low-reliability data has increased. do.

For example, in the update trigger determination unit 24, the average value of the reliability of the difficulty level stored in the difficulty level table stored in the parameter storage unit 32 or the average value of the reliability of the proficiency level stored in the proficiency level table is set. , When the ratio falls below the predetermined threshold L1 or when the ratio of the data below the predetermined threshold L2 to the whole increases, it is determined that the proficiency level and the difficulty level need to be updated, and the difficulty level proficiency level is determined. For the trigger, 1 is described in the ON / OFF column.

(2-3-4) When the estimated correct answer probability and the correct / incorrect data do not match The update trigger determination unit 24 determines the correct answer probability obtained based on the calculated parameters and the correct answer probability obtained from the actual correct / incorrect data. When it is determined that the divergence is large, it is determined that the proficiency level and the difficulty level need to be updated.

For example, the update trigger determination unit 24 may obtain the probability that the user j correctly answers the item i by substituting the proficiency level of each user and the difficulty level of each item stored in the parameter storage unit 32 into the equation 1. can. When this probability is estimated to be correct when it is 0.5 or more and incorrect when it is less than 0.5, the threshold value R for which the matching rate with the correct / incorrect data stored in the correct / incorrect data storage unit 31 is set in advance. If it is less than, the update trigger determination unit 24 determines that the proficiency level and the difficulty level need to be updated, and describes 1 in the ON / OFF column for the difficulty level proficiency level trigger. The threshold value R is a decimal number less than 1 such as 0.7.

(3) Parameter update processing Next, some possible examples of processing of each unit of the update control unit 26 that has received the update trigger from the trigger output control unit 25 will be described.

(3-1) Update of proficiency level (3-1-1) Update of proficiency level for a new user or an existing user who has new correct / incorrect data FIG. 6 shows a new user executed by the proficiency level update control unit 261. , Or, is a flowchart showing a processing procedure and processing content of proficiency level update for an existing user having new correct / incorrect data.

First, in step S201, the proficiency level update control unit 261 receives the proficiency level update trigger output from the trigger output control unit 25, and starts the following processing.

In step S202, the proficiency level update control unit 261 extracts the correct / incorrect data associated with the user ID notified together with the proficiency level update trigger from the correct / incorrect data storage unit 31 under the control of the data extraction unit 2611.

In step S203, the proficiency level update control unit 261 selects reference data to be used for proficiency level update from the correct / incorrect data extracted in step S202 under the control of the reference data selection unit 2612. In the selection, the reference data selection unit 2612 selects an item having a higher reliability from the item-specific difficulty parameters stored in the parameter storage unit 32 with the preset threshold value T3 as a reference.

In step S204, the proficiency level update control unit 261 calculates the proficiency level by using the items selected in step S203 above under the control of the proficiency level / reliability calculation unit 2613. For the calculation of the proficiency level, the difficulty level of the reference data can be substituted into Equation 2, and a maximum likelihood estimation method or another method can be used.

In step S205, the proficiency level update control unit 261 calculates the reliability of the proficiency level calculated in step S204 under the control of the proficiency level / reliability calculation unit 2613. As an index of reliability, the above-mentioned amount of test information or reliability using the Monte Carlo method can be used.

Subsequently, in step S206, the proficiency level update control unit 261 has high reliability of the proficiency level calculated in step S204 based on the reliability calculated in step S205 under the control of the reliability determination unit 2614. Judge. Here, if the calculated reliability is equal to or higher than the preset threshold value K1, the process proceeds to step S207.

In step S207, the proficiency level update control unit 261 outputs the calculated proficiency level and reliability to the parameter storage unit 32 and stores them.

On the other hand, if the calculated reliability in step S206 is less than the threshold value K1, the process proceeds to step S208. At this time, the reliability determination unit 2614 outputs to the reference data selection unit 2612 an instruction to change the threshold value T3 used for selecting the reference data and to reselect the reference data.

In step S208, the proficiency level update control unit 261 performs a process of lowering the threshold value T3 used for selecting the reference data to T3 ₁ in order to increase the reference data under the control of the reference data selection unit 2612 (T3>. T3 ₁ ).

Subsequently, returning to step S203, the proficiency level update control unit 261 reselects the reference data using the changed threshold value _T31 under the control of the reference data selection unit 2612.

Next, the proficiency level update control unit 261 re-executes steps S204 to S206 using the reselected reference data, and determines the reliability of the calculated new proficiency level. In step S206, the change of the threshold value T3 _x and the calculation of the parameter are repeated until the newly calculated reliability becomes the threshold value K1 or more. In step S206, when the reliability is still less than K1 and the threshold value T3 _x used for selecting the reference data becomes T3 _min preset as the minimum value thereof, the proficiency level update control unit 261 The process can be terminated without updating the proficiency level.

(3-1-2) Update of proficiency level for long-term users Even when new correct / incorrect data is obtained for existing users, especially long-term users, the same processing as in (3-1-1) above can be performed. can. However, if the user has been using the service for a long period of time, or if there is a rapid improvement, even if the period is not long, it is difficult to express the user's nature with a single proficiency level, and the model and actual data It is expected that the fit will be low. In order to avoid such a discrepancy between the model and the data, the data of the same user can be divided and regarded as a plurality of users having different proficiency levels.

FIG. 7 is a flowchart which is executed by the proficiency level update control unit 261 and shows a processing procedure and processing contents of the proficiency level update for a user whose proficiency level is significantly changed. This process may be performed automatically when, for example, correct / incorrect data having a specific user ID is accumulated for a certain period of time, or in response to an instruction input from the user or the operator. It may be carried out.

First, in step S301, the proficiency level update control unit 261 receives the proficiency level update trigger output from the trigger output control unit 25, and starts the following processing.

In step S302, the proficiency level update control unit 261 extracts the correct / incorrect data associated with the user ID notified together with the proficiency level update trigger from the correct / incorrect data storage unit 31 under the control of the data extraction unit 2611. At this time, the data extraction unit 2611 refers to the data ID used for calculating the proficiency level table stored in the parameter storage unit 32, and collects only correct / incorrect data that is not used for calculating the proficiency level currently stored. Can be obtained.

In step S303, the proficiency level update control unit 261 selects reference data to be used for proficiency level update from the correct / incorrect data extracted in step S302 under the control of the reference data selection unit 2612. In the selection, the reference data selection unit 2612 selects an item having a higher reliability from the item-specific difficulty parameters stored in the parameter storage unit 32 with the preset threshold value T4 as a reference. It is desirable that T4 is the same as T3, but different values may be used.

In step S304, the proficiency level update control unit 261 calculates the proficiency level by using the items selected in step S303 above under the control of the proficiency level / reliability calculation unit 2613. The same method as in (3-1-1) above can be used to calculate the proficiency level.

In step S305, the proficiency level update control unit 261 calculates the reliability of the proficiency level calculated in step S304 under the control of the proficiency level / reliability calculation unit 2613. The same method as in (3-1-1) above can also be used for calculating the reliability.

In step S306, the proficiency level update control unit 261, under the control of the reliability determination unit 2614, determines the high reliability of the proficiency level calculated in step S304 based on the reliability calculated in step S305. judge. Here, if the calculated reliability is equal to or higher than the preset threshold value K2, the process proceeds to step S307.

In step S307, the proficiency level update control unit 261 refers to the parameter storage unit 32 and calculates the difference between the proficiency level calculated up to the previous time and the proficiency level newly calculated in step S304.

In step S308, the proficiency level update control unit 261 determines that the change in proficiency level is large if the difference in proficiency level from the previous time is equal to or higher than the preset threshold value Q, and proceeds to step S309.

In step S309, the proficiency level update control unit 261 assigns a new user ID to the proficiency level calculated in step S304 as another user whose proficiency level is different, although the data is actually the same user, and stores parameters. Save to section 32. At this time, the proficiency level update control unit 261 deletes the correct / incorrect data used in step S302 from the correct / incorrect data storage unit 31, reattaches a new user ID, and saves the data. For example, as shown in FIG. 11, by assigning a new user ID "2_2", it can be distinguished from the original user ID "2". The value of K2 is preferably the same as that of K1, but different values can be used.

On the other hand, in step S308, when the difference in proficiency level from the previous time does not reach the preset threshold value Q, the process proceeds to step S311.

In step S311, the proficiency level update control unit 261 recalculates the proficiency level using all the data of the corresponding user stored in the correct / incorrect data storage unit 31.

Next, in step S312, the proficiency level update control unit 261 updates the recalculated proficiency level as a new proficiency level of the existing user, and stores it in the parameter storage unit 32.

On the other hand, if the calculated reliability in step S305 does not reach the threshold value K2, the process proceeds to step S310. At this time, the reliability determination unit 2614 outputs to the reference data selection unit 2612 an instruction to change the threshold value T4 used for selecting the reference data and to reselect the reference data.

In step S310, the proficiency level update control unit 261 performs a process of lowering the threshold value T4 used for selecting the reference data to T4 ₁ in order to increase the reference data under the control of the reference data selection unit 2612 (T4>. T4 ₁ ).

Subsequently, returning to step S303, the proficiency level update control unit 261 reselects the reference data using the changed threshold value _T41 under the control of the reference data selection unit 2612.

Next, the proficiency level update control unit 261 re-executes steps S304 to S306 using the reselected reference data, and determines the reliability of the calculated new proficiency level. In step S306, the change of the threshold value T4 _x and the calculation of the parameters are repeated until the newly calculated reliability becomes the threshold value K2 or more. In step S306, when the reliability is still less than K2 and the threshold value T4 _x used for selecting the reference data becomes T4 _min preset as the minimum value thereof, the proficiency level update control unit 261 sets the proficiency level update control unit 261. The process can be terminated without updating the proficiency level.

(3-1-3) Periodic proficiency update for long-term users As another example, correct / incorrect data is divided for each specific period L (for example, 1 week, 1 month, 3 months, etc.). It is also possible to use a method of newly allocating a user ID and calculating the proficiency level for each period L.

(3-2) Difficulty update (3-2-1) Difficulty update for new items or existing items with new correct / incorrect data Figure 8 shows the difficulty level update control unit 262, which is already difficult. It is a flowchart which shows the process procedure and process content which calculates and updates the difficulty degree for a new item or the item which has new correctness data with the same scale as the difficulty level scale of the item for which the degree is requested.

First, in step S401, the difficulty level update control unit 262 receives the difficulty level update trigger output from the trigger output control unit 25, and starts the following processing.

In step S402, the difficulty level update control unit 262 extracts the correct / incorrect data associated with the item ID notified together with the difficulty level update trigger from the correct / incorrect data storage unit 31 under the control of the data extraction unit 2621.

In step S403, the difficulty level update control unit 262 selects the reference data to be used for the difficulty level update from the correct / incorrect data extracted in step S402 under the control of the reference data selection unit 2622. In the selection, the reference data selection unit 2622 refers to the user who has data for the new item when the reliability of the proficiency level of the user stored in the parameter storage unit 32 exceeds the preset threshold value T5. Let it be data.

In step S404, the difficulty level update control unit 262 calculates the difficulty level of the corresponding item based on the data selected in the above step S403 under the control of the difficulty level / reliability calculation unit 2623. In this step, for example, the proficiency level of the reference data can be substituted into Equation 2, and the difficulty level can be obtained by a method such as maximum likelihood estimation.

In step S405, the difficulty level update control unit 262 calculates the reliability of the difficulty level calculated in step S404 under the control of the difficulty level / reliability calculation unit 2623 by using a method such as the amount of test information. ..

Subsequently, in step S406, the difficulty level update control unit 262 compares the reliability calculated in step S405 with the preset threshold value K3 under the control of the reliability determination unit 2624, and the reliability is K3 or higher. If so, the process proceeds to step S407.

In step S407, the difficulty level update control unit 262 outputs the calculated difficulty level and reliability to the parameter storage unit 32 and stores them.

On the other hand, if the calculated reliability in step S406 does not reach the threshold value K3, the process proceeds to step S408. At this time, the reliability determination unit 2624 outputs to the reference data selection unit 2622 an instruction to change the threshold value T5 used for selecting the reference data and to reselect the reference data.

In step S408, the difficulty level update control unit 262 performs a process of lowering the threshold value T5 used for selecting the reference data to T5 ₁ in order to increase the reference data under the control of the reference data selection unit 2622 (T5>. T5 ₁ ).

Subsequently, returning to step S403, the difficulty level update control unit 262 reselects the reference data using the changed threshold value _T51 under the control of the reference data selection unit 2622.

Next, the difficulty level update control unit 262 re-executes steps S404 to S406 using the reselected reference data, and determines the reliability of the calculated new proficiency level. In step S406, the change of the threshold value T5 _x and the calculation of the parameters are repeated until the newly calculated reliability becomes the threshold value K3 or higher. In step S406, when the reliability is still less than K3 and the threshold value T5 _x used for selecting the reference data becomes T5 _min preset as the minimum value thereof, the difficulty level update control unit 262 sets the difficulty level update control unit 262. The process can be terminated without updating the difficulty level.

(3-3) Simultaneous update of proficiency level and difficulty level The conventional method calculates the difficulty level from a small amount of data at the beginning, and keeps using the difficulty level fixedly. Therefore, when a large amount of data is accumulated, The model's fit to the entire data can be poor, especially if the user population changes. Simply re-estimating the proficiency level and difficulty level for new data using item response theory has a large difference from the previous difficulty level scale, and the user feels uncomfortable when compared with the previous parameters. It is supposed to be remembered.

Therefore, the learning support device 1 according to the embodiment performs a process of updating the difficulty level scale in the simultaneous update control unit 263 while reducing the discomfort of the user.

Upon receiving the simultaneous update trigger output from the trigger output control unit 25, the simultaneous update control unit 263 starts the update control process of the proficiency level and the difficulty level. At this time, the simultaneous update control unit 263 calculates a new proficiency level and difficulty level while referring to the data of the correct / incorrect data storage unit 31 and the parameter storage unit 32, and saves the calculation result in the parameter storage unit 32. As described above, the simultaneous update is not necessarily limited to the process of updating the proficiency level and the difficulty level at the same timing, but includes the process of updating the proficiency level and the difficulty level within a certain period of time.

A large number of examples can be considered for the specific processing performed by the simultaneous update control unit 263. Some of them are described below.

(3-3-1) Update Using Update Rate α FIG. 9 shows an example of a processing flow for simultaneously updating the difficulty level and the proficiency level, which is executed by the simultaneous update control unit 263.

First, in step S501, the simultaneous update control unit 263 receives the simultaneous update trigger output from the trigger output control unit 25, and starts the following processing.

In step S502, the simultaneous update control unit 263 selects reference data under the control of the reference data selection unit 2631. In the selection, the parameter storage unit 32 is referred to, and only the difficulty level whose reliability is the threshold value T6 or more and the proficiency level whose reliability is the threshold value T7 or more are selected, and the corresponding items and the user. The correct / incorrect data regarding the above is extracted from the correct / incorrect data storage unit 31. T6 is preferably the same as T3 used in (3-1-1) above, but may have a different value. Further, T7 is preferably the same as T5 used in (3-2-1) above, but may have a different value.

Next, in step S503, the simultaneous update control unit 263 calculates the proficiency level, the difficulty level, and the respective reliability levels using the selected reference data under the control of the proficiency level / difficulty level / reliability calculation unit 2632. .. Many methods are assumed as parameter calculation methods. For example, a method using all correct / incorrect data for a reference user and using a peripheral maximum likelihood estimation method, or a method in which the difficulty level of the reference data is substituted into Equation 2 is the most suitable. There is a method of calculating the proficiency level using the maximum likelihood estimation method, substituting the obtained proficiency level into Equation 2, and repeatedly calculating the difficulty level using the maximum likelihood estimation method, etc., until it converges. It is possible.

In step S504, the simultaneous update control unit 263 determines in the reliability determination unit 2633 that the reliability of each of the new proficiency level and the difficulty level calculated in this way is equal to or higher than the preset threshold value. After that, it is saved in the parameter storage unit 32.

としたときに、更新率αを用いて、それぞれ以下の式５および式６に従って更新することができる。

式中、ｓは何回目の更新であるかを表す整数であり、同時更新制御部２６３において難易度および習熟度の更新がなされ、ステップＳ５０４でパラメータ記憶部３２へと保存されるときに、ｓの値を１増やすようにすることができる。 Here, the new proficiency level and difficulty level calculated as above

Then, the update rate α can be used to update according to the following equations 5 and 6, respectively.

In the formula, s is an integer indicating the number of updates, and when the difficulty level and the proficiency level are updated by the simultaneous update control unit 263 and stored in the parameter storage unit 32 in step S504, s is used. The value of can be increased by 1.

For users who are not reference users, the proficiency level can be obtained by using the same method as the proficiency level calculation for new user data described in (3-1-1) above.

Further, if there is no problem in the service form even if there is a discrepancy with the difficulty level in the past, by setting α = 1, only the newly calculated parameter can be used.

(3-3-2) Update as time-series data As a second example of simultaneously updating the difficulty level and proficiency level, for example, when the difficulty level and proficiency level are calculated periodically, each parameter is time-series. It can be handled as data. At this time, in order to prevent the user from feeling uncomfortable with the change in the difficulty level, the difficulty level can be changed gently by using a method such as a moving average.

とする。このとき、パラメータ記憶部３２へ保存し、ユーザへのフィードバック等に使用するパラメータを

としたとき、これらは以下の式７および式８により算出される。

上記は、単純移動平均と呼ばれ、ｍは移動平均に用いる時点数を示す。単純移動平均以外にも、加重移動平均、指数移動平均、修正移動平均、三角移動平均、正弦加重移動平均、累積移動平均、ＫＺフィルタ、といった一般的に用いられている移動平均の手法を適応することもできる。 The proficiency level and difficulty level calculated from the data accumulated at time point t

And. At this time, the parameters stored in the parameter storage unit 32 and used for feedback to the user and the like are stored.

Then, these are calculated by the following equations 7 and 8.

The above is called a simple moving average, and m indicates the number of time points used for the moving average. In addition to the simple moving average, commonly used moving average methods such as weighted moving average, exponential moving average, modified moving average, triangular moving average, sine weighted moving average, cumulative moving average, and KZ filter are applied. You can also do it.

III. Effect As described in detail above, in the learning support device 1 according to the embodiment, the correct / incorrect data acquisition unit 21 acquires the correct / incorrect data and stores the correct / incorrect data in the correct / incorrect data storage unit 31. Then, various parameters are calculated by the parameter calculation unit 22 based on the correctness data and stored in the parameter storage unit 32. The stored parameters are read out and output by the parameter output control unit in order to present them to the user, the system administrator, or the like at any time, for example, in response to a request from an external application.

Here, the update trigger determination unit 24 and the trigger output control unit 25 monitor the data stored in the correct / incorrect data storage unit 31 and the parameter storage unit 32, or receive update instructions from inside and outside to update the parameters. It determines the necessity, generates a necessary update trigger, and notifies the update control unit 26. Upon receiving the update trigger, the update control unit 26 performs the update control process of the instructed parameter. At that time, the update control unit 26 extracts only highly reliable data from the data stored in the correct / incorrect data storage unit 31 and the parameter storage unit 32 and uses it as reference data to recalculate the parameters. I have to. Further, the data stored in the parameter storage unit 32 is updated only when the reliability of the recalculated parameter is higher than a certain value.

As described above, the problem of deterioration in proficiency level reliability due to variation in reliability level of difficulty described as (Problem 1) is solved. That is, even if an item is solved less frequently, or even if it is solved a lot but only correct answers or the correct answer rate is too low, the data corresponding to the unreliable difficulty level is excluded, and the highly reliable difficulty level is excluded. Since the new proficiency level is calculated based on the parameters, the reliability of the newly calculated proficiency level can be maintained. Further, since only the newly calculated proficiency level with high reliability is used for updating, the reliability of the parameters stored in the parameter storage unit 32 can be maintained.

In addition, the problem of deterioration in difficulty reliability due to variation in reliability of proficiency level described as (Problem 2) is also solved. That is, even if the number of answers by the user is small, or if the correct answer rate is too high or too low, the data corresponding to the unreliable proficiency level is excluded, and a new one is based on the highly reliable proficiency level parameter. Since the difficulty level is calculated, the reliability of the newly calculated difficulty level can be maintained. Further, since only the newly calculated difficulty level with high reliability is used for the update, the reliability of the parameter stored in the parameter storage unit 32 can be maintained.

If there are long-term users or users whose proficiency level has changed significantly, a new proficiency level is calculated using only the newly acquired correct / incorrect data, and the difference from the previous proficiency level is used. By assigning a new user ID to a new proficiency level, or by periodically dividing the data and assigning different user IDs, the problem of deterioration in the reliability of the proficiency level can be alleviated.

Furthermore, the problem described as (Problem 3) due to the small amount of data at the time of initial fixation of the difficulty scale is also necessary when the cumulative number of data items increases or the reliability decreases. It can be solved by updating to an estimated value based on more correct / incorrect data.

Furthermore, the user's discomfort is reduced by mitigating sudden parameter changes, such as weighting the newly calculated parameters and the previously calculated parameters using the update rate α and then adding them. It is possible to realize the parameter update. Similarly, by treating it as time-series data and taking a moving average to smooth out the change, it is possible to realize the update process that also reduces the user's discomfort.

[Other embodiments]
The present invention is not limited to the above embodiment. For example, the learning support device 1 is not limited to the one used in the online learning system, and can also be used for stand-alone learning. Further, as the data handled by the learning support device 1, the correct / incorrect data consisting of binary (correct / incorrect) responses to the items has been described, but the data is not limited to this.

The learning support device 1 may be implemented as a single device, but the functional units 21 to 26 included in the learning support device 1 may be distributed and arranged on a cloud server, an edge server, or the like, and these may cooperate with each other. The processing may be performed. Alternatively, the learning support device 1 may be provided as one of its extended functions on a user terminal or the like possessed by the user.

In addition, the format of the data stored in the correct / incorrect data table, the proficiency level table, the difficulty level table, the trigger table, and the like can be variously modified and implemented without departing from the gist of the present invention.

In short, the present invention is not limited to the above-described embodiment as it is, and at the implementation stage, the components can be modified and embodied within a range that does not deviate from the gist thereof. In addition, various inventions can be formed by an appropriate combination of the plurality of components disclosed in the above-described embodiment. For example, some components may be removed from all the components shown in the embodiments. In addition, components from different embodiments may be combined as appropriate.

1 ... Learning support device, 10 ... Input / output interface unit, 20 ... Control unit, 21 ... Correct / wrong data acquisition unit, 22 ... Parameter calculation unit, 23 ... Parameter output control unit, 24 ... Update trigger judgment unit, 25 ... Trigger output control Unit, 26 ... Proficiency update control unit, 30 ... Storage unit, 31 ... Correct / wrong data storage unit, 32 ... Parameter storage unit, 33 ... Trigger table storage unit,
261 ... Proficiency update control unit, 262 ... Difficulty update control unit, 263 ... Simultaneous update control unit, 2611 ... Data extraction unit, 2612 ... Reference data selection unit, 2613 ... Proficiency / reliability calculation unit, 2614 ... Reliability Judgment unit, 2621 ... Data extraction unit, 2622 ... Reference data selection unit, 2623 ... Difficulty / reliability calculation unit, 2624 ... Reliability judgment unit, 2631 ... Reference data selection unit, 2632 ... Proficiency / difficulty / reliability Calculation unit, 2633 ... Reliability determination unit.

Claims (8)

A learning support device that supports user learning
A correct / incorrect data acquisition unit that acquires correct / incorrect data indicating the correctness of each answer of a plurality of users for a plurality of items to be learned, and a correct / incorrect data acquisition unit.
Based on the acquired correctness data, the proficiency level indicating the high ability of each user, the difficulty level indicating the difficulty of each item, and the reliability of each of the proficiency level and the difficulty level. A parameter calculation unit that calculates the reliability that represents the height, and
A trigger generator that generates a trigger instructing to update at least one of the calculated proficiency levels or difficulty levels.
Update by the trigger by using the proficiency or difficulty having a reliability exceeding the first threshold value among the calculated proficiency or difficulty according to the generated trigger as a reference parameter. A learning support device including an update control unit that recalculates at least one of the instructed proficiency level or difficulty level. When the number of correct / incorrect data not used for calculating the proficiency level for a specific user exceeds the first number, the trigger generation unit receives an instruction to update the proficiency level for the specific user, or When the confidence level of the proficiency level calculated for the specific user is lower than the second threshold value and there is correct / incorrect data that is not used for calculating the proficiency level for the specific user, the proficiency level of the specific user is used. The learning support device according to claim 1, which generates a proficiency level update trigger instructing degree update. When the number of correct / incorrect data not used for calculating the difficulty level for a specific item exceeds the second number, the trigger generation unit receives an instruction to update the difficulty level for the specific item, or When the reliability of the difficulty calculated for the specific item is lower than the third threshold value and there is correct / incorrect data that is not used for calculating the difficulty for the specific item, the difficulty of the specific item The learning support device according to claim 1, which generates a difficulty level update trigger for instructing degree update. When the cumulative number of correct / incorrect data acquired by the correct / incorrect data acquisition unit reaches a natural number multiple of the third number, the trigger generation unit receives an instruction to simultaneously update the proficiency level and the difficulty level. When the average proficiency level for all users or the average difficulty level for all items is lower than the fourth threshold, the proficiency level for all users or all items It is calculated from the correct / incorrect data acquired when the ratio of the reliability lower than the fifth threshold value in the reliability of the difficulty level exceeds the sixth threshold value or by the correct / incorrect data acquisition unit. When the match rate between the correct answer rate and the correct answer rate calculated based on the predetermined relational expression based on the proficiency level and the difficulty level calculated by the parameter calculation unit is lower than the seventh threshold value. The learning support device according to claim 1, which generates a simultaneous update trigger instructing simultaneous update of the proficiency level and the difficulty level. When the trigger instructs the update of the proficiency level of a specific user, the update control unit has not yet been used to calculate the proficiency level among the correct / incorrect data having the user identification information associated with the specific user. A new proficiency level is calculated based on the correctness data, and if the difference between the new proficiency level and the previously calculated proficiency level is larger than the eighth threshold value, the new proficiency level and its calculation are used. The learning support device according to claim 1, wherein new user identification information different from the user identification information of the specific user is given to the used correct / incorrect data. The update control unit multiplies each of the recalculated proficiency level or difficulty level and the previously calculated proficiency level or difficulty level by a weighting coefficient and adds them, or the recalculated proficiency level or difficulty level. The learning support device according to claim 1, wherein the learning support device updates the proficiency level or the difficulty level by taking a moving average based on the degree and the proficiency level or the difficulty level previously calculated. It is a learning support method executed by a learning support device that supports the user's learning.
The process of acquiring correct / incorrect data indicating the correctness of each answer of multiple users for multiple items to be learned, and
Based on the acquired correctness data, the proficiency level indicating the high ability of each user, the difficulty level indicating the difficulty of each item, and the reliability of each of the proficiency level and the difficulty level. The process of calculating the reliability that represents the height, and
The process of generating a trigger instructing to update at least one of the calculated proficiency levels or difficulty levels, and
Update by the trigger by using the proficiency or difficulty having a reliability exceeding the first threshold value among the calculated proficiency or difficulty according to the generated trigger as a reference parameter. A learning support method comprising the process of recalculating at least one of the indicated proficiency levels or difficulty levels. A program for causing a processor to execute a process by each part of the learning support device according to any one of claims 1 to 6.

Images